HackAI Chrome Extension

Overview

The HackAI Chrome Extension is a lightweight tool designed to streamline AI model evaluation and feedback annotation directly in the browser. It serves as a key entry point for users to participate in adversarial testing workflows and contribute to human-in-the-loop safety validation.

Core Features

  • Prompt Testing Interface:Users can interact with models (starting with ChatGPT) inside the extension to test for prompt robustness, harmful content, and abnormal behavior.

  • Annotation and Feedback:The extension enables users to annotate outputs—flagging issues like hallucinations, bias, safety concerns, or incoherence.

  • Seamless Wallet Connection:Easily connect to supported wallets to authenticate user sessions and interact with the broader HackAI dApp ecosystem.

  • Local-First Execution:Prompt tests are executed locally in the browser. No prompts or outputs are uploaded unless the user explicitly chooses to submit feedback.

Product Context

The extension integrates with HackAI’s broader architecture, feeding annotated outputs into downstream modules such as:

  • Adversarial Sandbox (for reproducible attack validation)

  • Control Layer (for human-in-the-loop labeling)

  • Security Vault (for audit and traceability)

By embedding the plugin into the user workflow, HackAI lowers the participation threshold and makes safety research accessible and transparent.

Getting Started

  1. Visit https://hackai.ai and install the Chrome Extension (Coming Soon)

  2. Connect your wallet

  3. Start testing prompts inside the extension

  4. Annotate and optionally submit outputs

  5. View activity and updates from the Launch dApp


Note: During the Genesis phase, the prompt-test feature supports ChatGPT only. More models will be integrated in future updates.

Last updated